The lexicon of feeling offended

Abstract

The work investigates a neglected but widespread emotion: feeling offended. After overviewing previous research on offense, which considers feeling offended mostly as a damage to a person’s image, a survey study is presented aimed at deepening the definition of feeling offended and the description of its typical episodes. The definition emerging from the content analysis allows to code and group the more frequent words used in the corpus and to run a lexicometric analysis that describes the emotions connected to the feeling of offense, the individual traits of offender and victim, the causes of offense, and the types of judgment that most typically appear offensive. The applicability of the present work is related to the extraction of potential states of this emotion in other corpora where implicit or past intergroup conflicts are not completely solved, by also allowing a possible annotation of their seriousness in view of a potential process of reconciliation. 1

Full text

!
!
!
!
!
Symposium!on!Emotion!Modelling!and!
Detection!in!Social!Media!and!Online!
Interaction!
!
!
In!conjunction!with!the!2018!Convention!of!the!Society!for!the!Study!
of!Artificial!Intelligence!and!Simulation!of!Behaviour!(AISB!2018)!
!
!
!
5th!April!2018!
!

Emotions@AISB 2018 preface
Preface
Emotions@AISB 2018: Symposium on Emotion Modeling and Detection in Social
Media and Online Interaction
Message from the Chairs
Welcome to Emotions@AISB 2018, the Symposium on Emotion Modeling and Detection in
Social Media and Online Interaction. The symposium is part of AISB 2018, the convention on
Artificial Intelligence organized by the Society for the Study of Artificial Intelligence and Sim-
ulation of Behaviour. This symposium addresses the opportunities and challenges of emotion
modeling and tools for online interaction. This interest is motivated by the worldwide di↵usion
of social media, which has profoundly changed the way we communicate and access informa-
tion. Everyday, people interact with each other to share opinions about commercial products
on dedicated platforms, report their personal experiences on microblogging and social network-
ing sites, try to solve domain-specific problems through collaborative knowledge building and
sharing in online question and answering.
On one hand, such a pervasive use of online social media in computer-mediated communi-
cation is opening new challenges for social sciences and human-computer studies. Indeed, one
of the biggest drawbacks of communication through social media is to appropriately convey
and recognize sentiment through text: while display rules for emotions exist and are widely
accepted in traditional face-to-face interaction, people might not be prepared for e↵ectively
dealing with the barriers of social media to non-verbal communication. On the other hand,
user-generated content comprises an invaluable wealth of data, ready to be mined for training
predictive models. As such, microblogging and online interaction analysis are attracting the
interest of researchers and practitioners in NLP, machine learning, big data analysis. Indeed,
analyzing opinions and emotions conveyed by microposts can yield a competitive advantage for
businesses, and can serve to gain crucial insights about political sentiment and election results
or other social issues.
Emotions@AISB 2018 aims at fostering discussion around interdisciplinary research area at
the intersection between cognitive sciences, computational linguistics, and social computing.
We have invited three paper categories, full papers, short papers, and poster and demo papers,
to encourage submissions of contributions describing di↵erent stages of research on a↵ect recog-
nition in social media, ethical concerns, and applications, with a special focus on education,
entertainment, health, e-government, games, and hate speech monitoring. We are pleased to
present a collection of five papers discussing theoretical models, empirical studies, and tools.
All papers went through a thorough review process that involved at least three reviewers and
were evaluated based on their originality, quality, and relevance to the workshop. Furthermore,
we invited two key researchers with major contributions in this field to discuss their visions
and share the state of their research with the community in form of a keynote: Diana Maynard
(University of Sheffield, UK) delivering a talk on ”Twits, Twats and Twaddle. Analysis of hate
speech towards politicians in the GATE social media toolkit” and Pietro Cipresso (University of
Milan Sacro Cuore’, Italy) that will report on ”The role of a virtual body in modeling emotions
for Social Media and Online Interaction: The BodyPass project”.
We thank the members of our program committee and additional reviewers for their con-
structive reviews: Alessandro Ansani, Ruth Aylett, Francesco Barbieri, Pierpaolo Basile, Valerio
Basile, Erik Cambria, Chlo´e Clavel, Mihaela Cocea, Danilo Croce, Rossana Damiano, Celso De
Melo, Anna Esposito, Valentina Franzoni, Marco Guerini, Delia Irazu Hernandez Farias, Simona
Frenda, Emiliano Lorini, Saif Mohammad, Alessandro Moschitti, Marinella Paciello, Isabella
Poggi, Paolo Rosso, Diana Santos, Bj¨orn Schuller, Mohammad Soleymani, Khiet Truong, Carlo
i

Emotions@AISB 2018 preface
Strapparava, Enrico Zovato. In addition, we thank all authors for submitting interesting papers,
AISB and the local organizers for hosting us and for their exceptional support. Last but not
least, we heartily thank our invited speakers, Diana Maynard and Pietro Cipresso, for agreeing
to share their expertise on key topics of Emotions@AISB2018.
March 27, 2018 Emotions@AISB2018 Co-Chairs
Francesca D’Errico, University of
Roma-Tre, Italy
Floriana Grasso, University of
Liverpool, UK
Malvina Nissim, University of
Groningen, The Netherlands
Nicole Novielli, University of Bari
Aldo Moro, Italy
Viviana Patti, University of Turin,
Italy
ii

Emotions@AISB 2018 Table of Contents
Table of Contents
Music Emotion Capture: sonifying emotions in EEG data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
George Langroudi, Anna Jordanous and Ling Li
Associating Colours with Emotions Detected in Social Media Tweets . . . . . . . . . . . . . . . . . . . . . 5
Robert Harvey, Andrew Muncey and Neil Vaughan
The lexicon of feeling o↵ended ........................................................... 9
Francesca D’Errico and Isabella Poggi
An experiment with an o↵-the-shelf tool to identify emotions in students’ self-reported
accounts................................................................................. 16
Lubna Alharbi, Floriana Grasso and Phil Jimmieson
Prosocial words in social media discussions on hosting immigrants. Insights for
psychological and computational field. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Francesca D’Errico, Marinella Paciello and Matteo Amadei
1

The lexicon of feeling offended
Francesca D’Errico1, Isabella Poggi1
Abstract.
The work investigates a neglected but widespread emotion:
feeling offended. After overviewing previous research on
offense, which considers feeling offended mostly as a damage to
a person’s image, a survey study is presented aimed at deepening
the definition of feeling offended and the description of its
typical episodes. The definition emerging from the content
analysis allows to code and group the more frequent words used
in the corpus and to run a lexicometric analysis that describes the
emotions connected to the feeling of offense, the individual traits
of offender and victim, the causes of offense, and the types of
judgment that most typically appear offensive. The applicability
of the present work is related to the extraction of potential states
of this emotion in other corpora where implicit or past intergroup
conflicts are not completely solved, by also allowing a possible
annotation of their seriousness in view of a potential process of
reconciliation. 1
1 INTRODUCTION
Emotions are an alert device that wakens our attention any time
an adaptively very important goal of ours is, or is likely to be,
dramatically achieved or thwarted. The dimension of valence
that characterises all emotions, their being pleasant or unpleasant
ones, depends on their warning about either a goal achievement
or thwarting, and all negative emotions are in some sense healthy
in that the unpleasant feeling they cast over us, by warning that a
relevant goal is thwarted, may induce us to care that goal in the
future.
Among the most important goals of humans are those of positive
image and self-image, the goals of eliciting positive evaluations
about ourselves both by other people and oneselves, and the
“image” and “self-image” emotions that monitor these relevant
goals are those generally called “self-conscious” emotions
(Lewis, 1986), like the positive one of pride and the negative
ones of shame or humiliation.
This work deals with an emotion that has not been so often
investigated so far: feeling offended. A very unpleasant emotion
that we feel when our image and possibly our self-image are
dramatically disrupted by some event that in some way reveals a
low consideration of ourselves on the part of other people. After
overviewing previous research on offense (Section 2), we
present a survey study on the feeling of offense (Sect. 3), then by
exploiting lexicometric techniques, by analysing the words used
by our subjects in answering our survey we extract how the
1 Dept. Philosophy Communication and Visual Arts, Univ. Roma Tre,
Via Ostiense 234, Rome (Italy). Email:
{Francesca.derrico}@uniroma3.it;
{isabella.poggi}@uniroma3.it
concepts displayed more deeply describe the emotions connected
to the feeling of offense, the personal characters of offender and
victim, the causes of offense, and the types of judgment that
most typically appear offensive (Sect. 4).
2. Related work
Feeling offended is a complex emotional state that is often
modulated by personal factors like gender and self-esteem on the
basis of different expectations or causal attributions (internal vs
external); but it also involves relational factors that affect the
interpretation of the offense, since the person who is the cause of
the offense can be a relative, a friend, an acquaintance, a
colleague; and being offended by these people, with whom one
has more or less involving social relationships, may imply heavy
emotional costs. Within personal factors, self-esteem plays a
crucial role in the feeling of offense, since it can affect self-
relevant emotions like shame and pride (Brown and Marshall,
2001) – people with low self-esteem tend to feel shame more
than others; while gender mediates the feeling of offense,
especially in family contexts (Mosquera et al., 2002) – women
feel offended more than men.
The multidimensional factors that characterize this feeling have
been investigated in various psychological fields, from the
dynamic approach to social psychology. According to Zander
(1976), the feeling of offense is a deeply unpleasant emotion
which goes through three phases: 1. identification of the cause,
interpreted as an insult to an ideal value; 2. feeling of offense,
with its relative intensity related to the ‘expectations of
recognition’, and 3. reaction to the feeling of offense, modelled
according to socio-historical variables.
Within the socio-cognitive framework, Mosquera et al. (2002)
found that in the so-called “honour cultures” like Spain, as
opposed, for example, to the Netherlands, the prototypical case
of offence takes place in public and is referred to masculinity or
to female sexual morality. According to Cohen et al. (1996), the
higher the honour concern and the significance of the honourable
person, the strongest the emotional response to insults. In
cultures with a stronger code of honour, like Spain, the reactions
to the offence are more embarrassed than elsewhere, mostly in
relation to threats to family honour, so important for individual
self-esteem that when someone is offended one’s own self-
esteem can be damaged too. Further gender differences are
reported in emotional reactions of both shame and anger in case
the insults undermine the sexual dimension, especially with
Spanish women, for whom the sexual code of honor (sexual
shame) is stronger than for Duch women.
Interpersonal and intergroup elements are also central to the
feeling of offense in studies on forgiveness (Mc Cullough, 2000;
Paleari & Regalia, 2005): those who feel offended may feel
inferior in terms of perceived control (Baumeister et al.,2003)

and experience feelings of victimization or anger (McCullough,
2000), and this results in a need to restore their sense of power,
by also increasing power-seeking behavior (Foster & Rusbult,
1999). Such needs are welcome within a possible socio-
emotional reconciliation perspective in which the offender
attempts to undertake a "cycle of apology-forgiveness" (Nadler,
2002): the responsible person puts one’s own personal image in
the hands of the other person, at the risk of not being forgiven.
This presupposes a willingness to forgive as a result of a long-
term reconciliation path, for example where the transgressor
admits one’s responsibilities.
Studies on forgiveness and reconciliation provide a complex
framework where personality, ruminating tendencies, emotional
stability, empathy towards the transgressor can lead the offended
person to forgive, if offenses are explicit (e.g., betrayal, physical
and emotional humiliation).
Three problems in these studies are, first, somewhat circularity,
in that they define self-esteem as what is affected by offenses;
second, that they mainly investigate the emotional responses
triggered by explicit offenses, especially verbal insults, while
neglecting less direct and less explicit offenses; third, that they
almost exclusively focus on the offense as seen as a public blow
to image before other people.
To fill in these gaps requires to investigate the feeling of offense
in everyday contexts of interpersonal relationships, and to assess
if it is triggered only or typically by explicit communicative
actions like public discredit or insults, or whether even by more
implicit and indirect actions.
This has been recently done by Poggi & D’Errico (2018), who in
a survey study examined several autobiographic reports about
the feeling of offense. Besides providing a formal definition of
this emotion and finding out its most typical causes, they show
that it is not triggered only by blatant aggressive communicative
acts, by often by more indirect and implicit blows, like non-
communicative actions and even non-actions. Further, the
feeling of offense is not only felt in public contexts, but much
more – and more importantly – in affective relationships: the
offense tends to elicit high arousal emotions, such as anger,
pride, and revenge, in males and people with high self-esteem,
while it triggers low arousal emotions, such as bitterness, shame,
and even fear, in people with low self-esteem. Such emotions
tend to be felt by females more than males, both due to cultural
pushes for women to lower self-esteem, and due to their higher
tendency to care personal relationship, hence to suffer more for
their breaking.
3. Feeling offended: a survey study
The goal of our work is to investigate the feeling of offense
through direct self report of people. The focus of our analysis is
the definition and description of this emotional state in terms of a
socio-cognitive model (Castelfranchi, 2000; Miceli and
Castelfranchi, 2014; D’Errico and Poggi, 2016) which views
emotions as complex subjective states, entailing feelings,
physiological, expressive and motivational states, triggered when
a given cognitive configuration is represented in the mind of a
person: the beliefs about an occurring or imagined event,
attributions and evaluations about the self, as well as the goals at
stake and the goals triggered by the emotional syndrome.
3.1. Research questions
On the basis of subjects’ reports, we want to fulfil the following
research goals:
1. to provide a definition of the feeling of offense in
terms of its “mental ingredients”, that is, the beliefs and goals
represented in the mind of the person who is feeling that
emotion;
2. to explore the associated lexicon of ‘feeling offended’
emotional states by means of a lexicometric analysis that will
help is to describe the connected emotions, the individual traits
of offender and victim, the causes of offense, and the types of
judgment that most typically appear offensive.
3.2. Method
To investigate these issues, we submitted a semi-structured
online survey to a sample of 129 participants, mainly Italians,
balanced and composed by 61% women (n.79, vs. 50 males), age
31,2 (SD = 14,1), the majority with a high school bachelor
(54%) or a University degree (26%).
The survey included 14 close and 11 open questions, asking how
frequently the participants felt offended, for what reasons, who
offended them, and in what life domains (work, family,
friends…). To go deeply into the emotional experience of feeling
offended, participants were asked to report one case in which
they felt so, the specific reasons why they did, if they believed
the other intended to offend, their relationship with the other
before and after the offense, and what other emotions they
connected to the feeling of offense. We also asked if they
reminded of some case in which another wanted to offend them
but they did not feel offended, and if so, why they did not; and
conversely, if in some cases another person had felt offended by
them, but should not have felt so, and why. Finally participants
were asked to provide a definition of “feeling offended”.
Two types of analysis were conducted on the open questions:
first, a classical manual content analysis, then an automatic
analysis through lexicometric techniques.
4. Content analysis
The content analysis of the open answers to the questionnaire –
both the participants’ definitions and their description of
offensive events – first allowed us to outline a general definition
of “feeling offended” and to extract its mental ingredients.
4.1. Definition
Feeling offended is a negative emotion felt by a person A,
caused by either a communicative or a non-communicative act
by another person B that results into an aggression to A’s image,
since it explicitly points at or implicitly entails a negative
property of A: a property worth a negative evaluation of A by B
with respect to an evaluation criterion relevant for the image
which A wants to project, and shared with B. Being attributed
this negative property is seen as a true wound to A’s image, that
in some way implies a lack of respect for A (lack of care for
his/her image), and the aggression on the part of B is considered
unjust by A, who thinks s/he does not really deserve to be
attributed such a property. Actually A, though sharing the

evaluation criterion with B, may not share the same factual
knowledge: A and B share the value in terms of which facts can
be judged, but not the really occurred facts. The problem with A
– which definitely causes A to feel offended – is that B is
relevant for A, in that A has/had the goal of keeping a positive
social relationship with B. The whole fact results in subsequent
negative social emotions of A towards B, such as disappointment
and feeling betrayed by B, finally ending with a break in the
social relationship of A with B, and at the same time, a loss of
self-esteem for A.
4.2. Mental ingredients
Defining an emotion in some sense means to find out the
necessary conditions for a person to feel that emotion; so the
above definition of “feeling offended” can be translated into a
set of conditions: the beliefs and goals that are
necessary/sufficient for a person to feel offended. Among them,
three types of conditions can be distinguished (somehow like in
Searle’s (1969) analysis of speech acts): a) preparatory
conditions, b) essential conditions, and c) aggravating
conditions.
PREPARATORY CONDITIONS:
1. A has the goal of a positive image before B
2. A has the goal of a positive image before third parties
C
3. A has the goal of a positive self-image
4. A believes that property X is pertinent for his goal of
image before B or before third parties C
ESSENTIAL CONDITIONS:
1. B perfoms an Action A
Or else
2. B omits to perform Action A
3. A believes that this explicitly communicates or
indirectly implies
4. That B attributes a flaw X to A
5. X thwarts the image that A wants to project of himself
to B
And/or to third parties C
And/or to him/herself
6. A believes that X makes him/her inferior to B / C
Or
7. to the category to which A wants others to believe he
belongs to.
8. All of this causes A to feel
a. a negative image emotion
(sadness, displeasure, shame, humiliation)
and/or
b. a negative social emotion towards B
(anger or rancour)
c. a negative emotion of affiliation
(inferiority, feeling of exclusion)
d. a negative emotion of attachment towards B
(disappointment about B)
AGGRAVATING CONDITIONS:
The negative emotion of A is as more dramatic as
1. The manifestation of A’s flaw is public, i.e.,
A believes that third parties C will come to know
about As’ flaw or inferiority
2. A believes that B’s attack to A’s image is deliberate
3. A has a low self-esteem
4. A’s self-image is strongly dependent on the image that
others (B and/or C) have of A
5. For A the goal of having a positive social
(possibly affective) relationship to B is important
6. A esteems B
4.3. Close emotions
One of the questions in the survey inquired what emotions are
closer to the feeling of offense, either in the sense of making part
of the feeling itself, or of being felt at the same time.
Participants in their answers would mention humiliation (12
occurrences), and particularly often anger (50), both along with
connected emotions like disappointment (29), and with
antagonistic ones like sadness (25), while only one case shame is
reported. Further mentioned emotions are bitterness (15) and
rancour (6); and the latter is often seen, on the one side, as a final
result of feeling offended, on the other as the reason why the
offense may finally cause a break of the social relationship
between A and B.
4.4. Causes of offense
Unlike the traditionally accepted view that people are most
typically offended by insults or other overt communicative acts
of discredit, the causes of offense resulting from our data show a
much wider range of events (Table 1).
a.
B’S
COMMUNICATIVE
ACTION
b.
B’S
ACTION
c.
B’S
IMPLIED MENTAL
STATE
criticism
omission
carelessness
accusation
attitude
unjust accusation
behavior
formal negative
judgment
arrogance
reproach
taking
advantage
insult
betrayal
preference
aggressiveness
(including
impoliteness)
incomprehensio
n
exclusion
distrust
negative
prediction
deception,
hypocrisy
slander
injustice
gossip
mockery
silence
refusal (of an

offer, an
invitation, of
listening)
Table 1 | Causes of offense
Actually, people can feel offended not only by overt
communicative actions, but also by non-communicative actions,
and even by non-actions: in a word, by anything that in some
way uncovers either a clearly stated or even an implied mental
state from which a low opinion of B about A can leak out.
COMMUNICATIVE ACTIONS (col. a)
A person can be offended by a criticism, a slander, an unjust
accusation; by gossip, insults, mockery, but also by a reproach, a
formal negative judgment (like a bad score); or by a negative
prediction, like in this example:
(Participant 107): quando a venti anni mi dicevano che non
avrei fatto molto nella vita
(when at 20 people told me I would not do so much in my life)
Yet, the bulk of offensive action is exclusion
(29): quando un professore mi ha cacciato da un esame orale
(when a teacher sent me away of an oral examination)
And exclusion can simply be communicated by a refusal.
NON-COMMUNICATIVE ACTIONS (col. b).
Sometimes, what is offensive for A is not a particular
communicative behavior, but a general attitude of B:
(35): Quando frequentavo l'università, una mia insegna[n]te,
nonché relatrice, spesso mi faceva sentire un’ignorante.
(As I attended the university, a teacher of mine, and tutor,
often made me feel an ignorant person)
One more offensive behavior is B’s taking advantage of A: this
makes A feel “used” like an object, not credited the dignity of a
human person with her personal goals and desires.
Finally, injustice is offensive: as stated in this case.
(27): A lavoro quando non mi è stato riconosciuto il merito di
un compito svolto
(At work when I was not acknowledged the merits of a task
performed)
Being subject to injustice is offensive also for an underlying
thought: how unworthy am I so as to be treated this way?
IMPLIED MENTAL STATE (col. c)
Sometimes A is offended not by what B does or does not do, but
by an implicit mental state of B that can be indirectly inferred
from B’s communicative or non-communicative behavior (col.
c). See this example:
(16): Quando ho dato dei consigli a dei familiari ma non mi
hanno ascoltato e si [sono] fidati di altri, i quali hanno fornito
le mie stesse opinioni.
(when I gave advice to relatives but they did not listen to me
and trusted others, who provided the same opinions)
Here what is offensive for A is a substantive distrust of B for A
that is made explicit by B’s not following A’s advice.
One more offensive mental state, generally implied by an
omission, is the other’s carelessness:
(52): Una mia cugina non ha mantenuto la sua promessa di
venirmi a trovare, e non mi ha più cercato
(A cousin of mine did not keep her promise to come visit me,
nor did look for me anymore)
That the other disregards her own promise means that you are
not important for her, she does not care you and your feelings:
something highly upsetting. Further, if your low importance for
the other is a bad hit to your self-esteem, even more so is the
comparison between how important you are for the other as
opposed to other people. Thus when the other prefers someone
else over you, you feel betrayed: and betrayal is not only
offensive per se but mostly because A finally loses in the
comparison between him/her and the rival, who is preferred by
B. Like in this example:
(112): mia sorella si è sposata e non mi ha voluto come
testimone dopo che me l'aveva già chiesto
(my sister got married and did not want me as her wedding-
witness, after she had asked me to)
In the last two cases B’s preference uncovers A’s relative
unimportance. Yet, the extreme case of this is not being
acknowledged at all as a person, for example, when the other
does not greet you when meeting you.
4.5. Offensive judgements
One more result emerging from the content analysis of open
questions is an overview of what kinds of judgments are
considered offensive. A judgment can be defined as an
evaluation, that is, a belief about how we match to some
standard, to some criterion of evaluation – in terms of our model,
how adequate we are with respect to some goal, that is, how
much power we have to achieve them – out of all the possible
evaluations we elicit from others, the ones that are offensive for
us are only those concerning those standards, those evaluation
criteria, that we deem as relevant to the image we want to
present to others and/or ourselves. But there are some criteria of
evaluation that are generally important for everybody, and
negative evaluations with respect to them are considered
offensive. Based on previous works on the discrediting acts in
political communication (Poggi et al., 2011; D’Errico & Poggi,
2012) we classified the offensive evaluations reported by our
participants as four kinds of inadequacy (Table 2): a. physical
(aesthetic or functional) inadequacy; b. competence (cognitive
skill like knowledgebility, planning and reasoning); c.
dominance (power, decisional effectiveness social influence
skills); and d. benevolence (ethical qualities, altruism, honesty,
morality).
Criterion of
evaluation
Negative judgement
PHYSICAL
aesthetic inadequacy (physical
appearance: ugly, dirty, ridiculous)
functional inadequacy (disability

and sexual inadequacy)
COMPETENCE
stupidity
ignorance
lack of skills
lack of social skills (e.g., choosing a
wrong boy-friend, wrong education
of kids)
DOMINANCE
impotence
affective irrelevance (e.g., not being
respected)
unusefulness
dependence (e.g. a
personality trait:
subjective)
inferiority (by status:
objective)
BENEVOLENCE
immorality
injustice
cheating
unreliability
negligence
egoism
Table 2 | Offensive judgements
A case in which people typically feel offended is when being
teased or criticized for their PHYSICAL APPEARANCE, based
on an aesthetic criterion: fat or clumsy are frequent offensive
epithets, while a more “creative” one is penguin – an indirect
way to name the same properties, reminded by a participant.
Even more than aesthetic inadequacy, targeting the functional
properties of a person’s physical arrangement is offensive: the
stygma of handicap humiliates people, hence being offensive,
even when the intention of B is not to offend but, for instance, to
help: pity is offensive and humiliating. Yet, no examples of this
kind of inadequacy are mentioned in our corpus.
Vis à vis the criterion of COMPETENCE, participants mention
stupidity, but also lack of social skills (like being told you are
not able to educate your children).
Concerning DOMINANCE, the other’s carelessness is offensive
since it tells they consider you irrelevant. You feel totally
inconsequential when you are considered or explicitly accused to
be useless, but also when people ignore you, or they omit those
simple acts that credit you dignity and consider you worth of
respect. Further offensive attacks to a person’s DOMINANCE
are when others consider them inferior.
Coming to BENEVOLENCE, people are offended by
accusations of immorality, of being a liar or an unreliable
person, of negligence or non-compliance with one’s duty, and
finally of selfishness.
5. Lexicometric analysis
The automatic quanti-qualitative analysis was performed on the
subjects’ answers by TalTac (Trattamento Automatico
Lessicale e Testuale per l’Analisi del Contenuto, i.e.“Lexical
and Textual Automatic Processing for Content Analysis”:
Bolasco, 2016), a software for textual data analysis based on a
“lexicometric approach”: an application of statistical principles
to textual corpora. The “textual statistics” aims to extract the
semantic level in a text starting from the list of words obtained
by statistical analysis; for example, in the analysis of
specificities, the software extracts a list of significant words
obtained by a statistical comparison between sub-parts of text
according to selected variables.
TalTac allows to extract the ‘peculiar lexicon’ (Bolasco, 2016)
that is, the set of the words over-represented or under-
represented in the text under analysis, by comparing the corpus
to an external frequency lexicon, taken as a reference model (in
our case, Standard Italian). The measure of the variance from the
reference lexicon is represented by the standard deviation, which
is the deviation between the form frequencies in the analysed
text and in the frequency lexicon.
The corpus extracted from all answers in our survey is
particularly feasible to lexicometric analysis since it allows to
extract cognitive elements of emotions through the words used
by people (Poggi & D’Errico, 2010; D’Errico & Poggi, 2014).
This present corpus counts 14061 (N) occurrences with 3003 (V)
different words and a medium lexical richness index
[(V/N)*100], equal to 21,35%.
To go more into the categories emerging from our content
analysis, we run a lexicometric analysis on the whole corpus of
our participants’ answers. This allowed us to single out all the
words fitting the three categories above, with their relative
frequencies: Emotions (emotional states), Actions and
Judgements. In addition another interesting category emerged:
one of Stable traits that may define either the Offender (O) or the
Victim, i.e., the Offended Person (V), and that may be the cause
or the condition of the feeling of offense or of the other
emotional state.
We coded a total of 492 words by also reading at the real context
of use (the actual sentences in which they occur and then
grouped considering the semantic similarity) and we found that
in our corpus the great part of this lexicon is related to the
description of the emotional states (48%), followed by the
actions that caused them (29%) the judgments (15%) and finally
the stable traits of Offender or Victim that may have caused or
allowed the emotional states.
ACTIONS TRAITS EMOTIONS JUDGMENTS
29% 8% 48% 15%
The frequency configuration of the most frequent words coded
as Emotions is summarized in Figure 1.
Figure 1. Emotions caused by the offense

From Figure 1 it emerges how feeling offended can be a negative
state associated mainly with low arousal emotions like delusa
(disappointed), triste (sad), ferita (wounded), rimanerci male
(take it bad), amareggiata (embittered), rancore (rancor),
umiliato (humiliated), dispiaciuta (sorry), scoraggiata
(discouraged), mortificata (mortified), imbarazzata
(embarrassed), sofferenza (suffering); but it also looks
connected, although to a lesser extent, with words mentioning
high arousal emotional states, such as arrabbiata (angry),
tensione (tension), frustrate (frustrated). These opposite
reactions to the feeling of offense can be linked to different
levels of individuals’ self-esteem, in that a passive reaction to an
offence is mainly linked to low levels of self-esteem (Poggi &
D’Errico, 2018).
As to the actions mentioned in the corpus, it emerges (Fig. 2)
that the offense can be mainly caused by negative judgments
expressed in different ways, ranging from accusato (accused),
giudicato (judged), rimproverata (reproached), rimprovero
(reproach), accusata (accused), critica (criticism), denigrato
(denigrated), deriso (teased), insult (insults), to different
judgments implied by provocation or jokes: ridere (laugh),
scherzava, scherzare (joke), battuta (humorous joke), frecciatina
(sideswipe), that can be a more indirect evaluation. Words that
are linked to causes of feeling offended can be related to a lack
of consideration, in this case people feel offended because they
are lasciato (forsaken), ignorato, ignorandola (ignored),
sminuito (diminished), allontanato (distanced), isolato (isolated).
A less common cause of offense are relational problems like
tradito (betrayed), incomprensione (incomprehension), frainteso
(misunderstood), or aggressive actions such as harsh and
impolite manners: scortesi (impolite), urlato (shouted), strillato
(screamed), in faccia (in your face). Yet, as is clear from the
high frequency showing in Fig.2, a word recurring very often is
tradito (betrayed): a very serious and heavy action. This again is
consistent with the above observation about the word amicizia
(friendship): what most typically enhances the suffering of
feeling offended is that a person to whom we feel bounded
unexpectedly violated our expectation about us, betraying what
we feel as a previous affective commitment s/he had with us.
Besides emotions, in the corpus amicizia (friendship) has a good
frequency among the words used by participants; this can be
accounted for by the fact that, as clearly demonstrated in the
whole work (Poggi & D’Errico, 2018), the most serious cause of
offense is when the offensive action comes from persons with
whom the Victim has an important affective relationship: we are
more offended by friends than by strangers.
Figure 2. Actions causing the offense
The category of Judgments (Fig. 3) collects all the negative
evaluations mentioned in the corpus that include lack of
benevolence – falso (false), cattiveria (badness), bugiarda (liar),
vigliacca (coward), cattive (bad), opportunismo (opportunism),
egoismo (selfishness) – lack of dominance – sciocchezze
(nonsense), debole (weak), inutile (useless), cazzata (crap),
inadatto (unfit), mediocrità (mediocrity), passiva (passive),
incapace (unable), – negative physical features – peso (weight),
aspetto (look), brutte, brutto (ugly), fisicità (physical), difetti
(flaws), complessate (full of complexes), bassezza (shortness),
ingrassata (fatter), grassa (fat), estetico (aesthetic), robusta
(stout) – and to a lesser extent lack of competence – stupida
(stupid), incompetente (incompetent).
Figure 3. Offensive judgments
A last category, not considered in the preceding work, that pops
up from lexical frequencies, includes adjectives or nouns
mentioning mental properties of person, that may be either the
Victim or the Offender. Examples of the first kind are
suscettibilità (hypersensitiveness), permalosa (touchy),
vulnerabili (vulnerable) and orgogliosa (proud). As resulted
from the content analysis by Poggi and D’Errico (2018), a
particularly touchiness of the Victim, possibly favored by a low
self-esteem, makes him or her more vulnerable to feeling
offended. At the same time, one who is orgogliosa (proud)
considers the goal of being appreciated and well judged by
others particularly important, hence lowering the threshold of
feeling offended. On the other hand, words like superior
(superior) superiorità (superiority) and presuntuosa
(presumptuous) describe the Offender, who tends to offend other
people because displaying or boasting one’s superiority, so much
so as to be seen (by the Victim) as presumptuous.

Figure 4. Offensive Traits
3.6. Discussion and conclusion
This work was aimed at investigating a neglected but widespread
emotion: feeling offended. After overviewing previous research
on offense, who considered feeling offended mostly as a blow to
the person’s public damage, we have presented a survey study on
the feeling of offense, aimed at deepening the very definition and
the description of feeling offended. The definition emerging
from the content analysis allowed us to code and group the more
frequent words used in the corpus and, though a lexicometric
analysis, to describe the emotions connected to the feeling of
offense, the individual traits of offender and victim, the causes of
offense, and the types of judgment that most typically appear
offensive. Beside working as a description of this emotion, the
present work might be further applied to detect the feeling of
offense in other corpora, and thus to discover implicit or past
conflicts not completely solved, by allowing also a possible
annotation of their seriousness. The automatic extraction of a
generalized online feelings of offence in intergroup conflicts, for
example, can be used to project a more emotionally focused
reconciliation where ‘offended group’ need to be empowered by
restabilising their psychological sense of control (Nadler, 2002).
In this sense within social media (Carr, 2017) the emotional
analysis can contribute to turn different forms of opinions
polarization in a deep understanding of offended people’s needs.
REFERENCES
[1] Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D.
(2003). Does high self-esteem cause better performance,
interpersonal success, happiness, or healthier lifestyles?.
Psychological science in the public interest, 4(1), 1-44.
[2] Bolasco S. (2016). L’analisi multidimensionale dei testi. Carocci
Roma
[3] Brown, J. D., & Marshall, M. A. (2001). Self-esteem and emotion:
Some thoughts about feelings. Personality and Social Psychology
Bulletin, 27(5), 575-584.
[4] Carr, C. (2017). Social Media and Intergroup Communication.
Oxford Research Encyclopedia of Communication.
[5] Castelfranchi, C. (2000). Affective Appraisal versus Cognitive
Evaluation in Social Emotions and Interactions. In Paiva, A. (ed.),
Affective Interactions. Towards a New Generation of Computer
Interfaces. Berlin: Springer, pp.76-106.
[6] Cohen, D., Nisbett, R. E., Bowdle, B. F., & Schwarz, N. (1996).
Insult, aggression, and the southern culture of honor: An"
experimental ethnography.". Journal of personality and social
psychology, 70(5), 945.
[7] D’Errico, F., & Poggi, I. (2012). Blame the opponent! Effects of
multimodal discrediting moves in public debates. Cognitive
Computation, 4(4), 460-476.
[8] D’Errico, F., & Poggi, I. (2014). Acidity. The hidden face of
conflictual and stressful situations. Cognitive Computation, 6(4),
661-676.
[9] D’Errico, F., & Poggi, I. (2016). Social emotions. A challenge for
sentiment analysis and user models. In Emotions and Personality in
Personalized Services (pp. 13-34). Springer International Publishing.
[10] Foster, C. A., & Rusbult, C. E. (1999). Injustice and powerseeking.
Personality and Social Psychology Bulletin, 25(7), 834-849.
[11] Lewis, M. (2008). Self-conscious emotions: Embarrassment, pride,
shame, and guilt. In M. Lewis, J. Haviland-Jones, & L. Feldman
Barrett (Eds.), Handbook of emotions, 3rd ed. (pp. 742-756). New
York:Guilford Press.
[12] McCullough, M. E. (2000). Forgiveness as human strength: Theory,
measurement, and links to well-being. Journal of Social and Clinical
Psychology, 19(1), 43-55.
[13] Miceli, M., & Castelfranchi, C. (2014). Expectancy and Emotions.
Oxford: Oxford University Press.
[14] Mosquera, P. M. R., Manstead, A. S., & Fischer, A. H. (2002). The
role of honour concerns in emotional reactions to offences. Cognition
& Emotion, 16(1), 143-163.
[15] Nadler, A. (2002). Social-psychological analysis of reconciliation:
Instrumental and socio-emotional routes to reconciliation. Peace
education worldwide: The concept, underlying principles, the
research, 127-143.
[16] Paleari, F. G., Regalia, C., & Fincham, F. (2005). Marital quality,
forgiveness, empathy, and rumination: A longitudinal analysis.
Personality and Social Psychology Bulletin, 31(3), 368-378.
[17] Poggi, I., & D’Errico, F. (2010). The mental ingredients of
Bitterness. Journal on multimodal user interfaces, 3(1), 79-86.
[18] Poggi, I., D’Errico, F., & Vincze, L. (2011). Discrediting moves in
political debate. In Proceedings of Second International Workshop on
User Models for Motivational Systems: the affective and the rational
routes to persuasion (UMMS 2011) (Girona) Springer LNCS (pp. 84-
99).
[19] Poggi, I., & D’Errico, F. (2018). Feeling offended. A blow to our
image and our social relationship. Vol. 8, p.2221, Frontiers in
Psychology. https://doi.org/10.3389/fpsyg.2017.02221
[20] Zander, W. (1976). Taking offence as seen in depth-psychology.
Psychotherapie und Medizinische Psychologie.